As is well known, a klystron oscillator of the type referred to includes electrodes (i.e. a cathode and an anode) between which an electron beam successively traverses the two cavities, passing through a buncher gap in the input cavity and then through a catcher gap in the output cavity. The electric field set up across the buncher gap modulates the velocity of the beam electrons which then pass through a drift space into the catcher gap where the resulting density variations give rise to electromagnetic oscillations fed back to the buncher gap. The oscillating frequency is determined by the dimensions of the two resonant cavities but, generally, is also subject to some variation in response to changes of the d-c biasing voltage across the electron-emitting cathode and the electron-collecting anode. This voltage dependence of the oscillator frequency is referred to in the art as "frequency pushing".
The frequency stability of such an oscillator is a function of the quality or Q factor of the output cavity and also varies generally inversely with the length of the drift space. To increase the Q factor, and thus to minimize the pushing effect and the attendant noise, it has already been proposed to couple a further resonant cavity to the output cavity or to insert such an additional cavity in the feedback path between the input and output cavities. These prior solutions of the problem of frequency stabilization, however, greatly complicate the structure of the klystron and increase its overall dimensions as well as its cost.